X (Twitter)

Docs

Embeddings

Generate vector embeddings for text using the embeddings endpoint.

The embeddings endpoint converts text into numerical vectors that capture semantic meaning. Use embeddings for similarity search, clustering, and RAG applications.

Endpoint

POST https://api.elyxir.ai/v1/embeddings

Request

Headers

Header	Required	Description
Authorization	Yes	Bearer token: `Bearer elyxir_xxx`
Content-Type	Yes	Must be `application/json`

Body Parameters

Parameter	Type	Required	Description
model	string	Yes	Embedding model ID
input	string/array	Yes	Text to embed
encoding_format	string	No	`float` (default) or `base64`

Example Request

curl -X POST https://api.elyxir.ai/v1/embeddings \
  -H "Authorization: Bearer elyxir_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "text-embedding-3-small",
    "input": "The quick brown fox jumps over the lazy dog"
  }'

Response

Success (200 OK)

{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "index": 0,
      "embedding": [0.0023, -0.0093, 0.0152, ...]
    }
  ],
  "model": "text-embedding-3-small",
  "usage": {
    "prompt_tokens": 9,
    "total_tokens": 9
  }
}

Batch Embeddings

Embed multiple texts in one request:

response = requests.post(
    "https://api.elyxir.ai/v1/embeddings",
    headers=headers,
    json={
        "model": "text-embedding-3-small",
        "input": [
            "First document to embed",
            "Second document to embed",
            "Third document to embed"
        ]
    }
)
 
# Response includes embedding for each input
for item in response.json()["data"]:
    print(f"Index {item['index']}: {len(item['embedding'])} dimensions")

Available Models

Model	Provider	Dimensions	Max Input
text-embedding-3-large	OpenAI	3072	8191
text-embedding-3-small	OpenAI	1536	8191
text-embedding-ada-002	OpenAI	1536	8191
embed-english-v3.0	Cohere	1024	512
embed-multilingual-v3.0	Cohere	1024	512

Use Cases

Semantic Search

Find similar documents:

import numpy as np
 
def cosine_similarity(a, b):
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))
 
# Embed query and documents
query_embedding = get_embedding("How do I reset my password?")
doc_embeddings = [get_embedding(doc) for doc in documents]
 
# Find most similar
similarities = [cosine_similarity(query_embedding, doc) for doc in doc_embeddings]
most_similar_idx = np.argmax(similarities)

RAG (Retrieval-Augmented Generation)

Embed your documents and store in a vector database
Embed the user's query
Find similar documents using vector similarity
Include relevant documents in the LLM prompt

Clustering

Group similar texts:

from sklearn.cluster import KMeans
 
# Get embeddings for all texts
embeddings = [get_embedding(text) for text in texts]
 
# Cluster into groups
kmeans = KMeans(n_clusters=5)
clusters = kmeans.fit_predict(embeddings)

Best Practices

Text Length

Shorter texts (1-2 sentences) work well for questions/queries
Longer texts (paragraphs) capture more context
Very long texts may exceed model limits

Preprocessing

Clean text of irrelevant formatting
Consider chunking long documents
Normalize text (lowercase, remove special characters) if needed

Choosing a Model

Use Case	Recommended Model
English text, high quality	text-embedding-3-large
English text, cost-effective	text-embedding-3-small
Multilingual	embed-multilingual-v3.0
Legacy compatibility	text-embedding-ada-002

API Overview - Getting started
Chat Completions - Text generation
Model Hub - Browse embedding models

Completions Error Handling